AI Terminology
/Intermediate
Precision
Definition
An evaluation metric that answers: 'Out of all the times the AI predicted Positive, how many were ACTUALLY Positive?' It measures the quality of a positive prediction.
Explain Like I'm New
The boy who cried wolf. If the AI sounds the alarm that there is a wolf, how much can you trust it? High Precision means when it sounds the alarm, there is almost certainly a wolf. Low Precision means it sounds the alarm constantly for no reason (False Positives).
Real World Example
YouTube's Copyright Strike AI. If it has Low Precision, millions of innocent creators get their videos taken down falsely. To avoid angry creators, the AI must be tuned for High Precision (only striking when it is 100% sure).
Common Use Cases
- •Spam filtering
- •Criminal conviction AI
Interview Questions
basic
- If Precision is low, what kind of errors is the AI making too often: False Positives or False Negatives?
intermediate
- What is the mathematical formula for Precision?